Bridge incremental construction multi-parameter monitoring system and method based on fusion of vision and laser

Abstract

The bridge incremental launching construction multi-parameter monitoring system and method based on fusion of vision and laser belong to the technical field of intelligent construction monitoring. By fusing AI video recognition, laser precise ranging and rigid body space transformation technology, a technical system of multi-source sensing fusion perception-rigid body kinematics intelligent solution-linear deviation differential analysis-hierarchical intelligent early warning is constructed. The present application realizes all-weather non-contact collaborative monitoring of steel support three-dimensional displacement, beam body central axis deviation and guide beam high-frequency dynamic deflection. The central axis deviation monitoring precision reaches ±1mm / 100m, the guide beam deflection monitoring frequency is up to 100Hz, the efficiency is improved by more than 10 times compared with traditional manual measurement, the safety hazard of personnel entering the high-risk operation area is completely avoided, and the present application is suitable for linear and curved incremental launching construction monitoring of various bridge steel-concrete composite beams and steel box beams.

Description

Technical Field

[0001] This invention belongs to the field of intelligent construction monitoring technology, specifically involving a multi-parameter monitoring system and method for bridge jacking construction based on vision and laser fusion. It is particularly suitable for monitoring the straight and curved jacking construction of steel-concrete composite beams and steel box girders in projects such as overpasses, river-crossing bridges, and interchange hubs. Background Technology

[0002] With the rapid development of national transportation infrastructure construction, bridge projects are increasingly crossing complex environments such as existing railways, highways, rivers, and canyons. The incremental launching method, due to its advantages such as not affecting traffic below the bridge, requiring no large-scale water or high-altitude work equipment, and easy control of construction quality, has been widely used in the construction of steel bridges and steel-concrete composite bridges. During the incremental launching process, the stress deformation of the temporary steel supports, the deviation of the central axis of the launched beam, and the dynamic deflection of the guide beam (guide beam) are key parameters directly related to construction safety and the quality of the completed bridge's alignment, and must be closely monitored throughout the entire process.

[0003] Currently, bridge jacking construction monitoring mainly employs two types of technical methods: manual measurement and automated monitoring using single sensors. However, these methods have the following limitations when addressing the high precision, high safety, and continuous all-weather operation requirements of modern bridge jacking construction: First, traditional manual measurement methods are inefficient and pose significant safety risks. Monitoring steel support deformation and centerline deviation mainly relies on total stations combined with manual point-by-point measurements. A single full-line measurement can take several hours, making it difficult to meet the real-time requirements of jacking construction. Surveyors need to frequently enter high-risk working environments such as under the jacking beam and in the support area, posing safety risks such as falls from heights and being struck by objects. Furthermore, manual monitoring is almost impossible to implement at night, in rain, snow, or heavy fog, resulting in blind spots in safety monitoring during construction intervals.

[0004] Secondly, automated monitoring using a single sensor suffers from insufficient accuracy and poor data coordination. Some projects have introduced static levels, inclinometers, or GPS for monitoring, but static levels can only monitor vertical settlement and cannot acquire horizontal displacement; GPS signals are affected by the bridge site environment (such as under the bridge or in a valley) and multipath effects, making it difficult to meet the millimeter-level measurement accuracy requirements; single video monitoring is easily affected by lighting and weather interference and lacks an absolute coordinate reference; more importantly, steel support deformation, axis deviation, and guide beam deflection are usually monitored independently by different systems, resulting in isolated data that cannot achieve comprehensive analysis and coordinated early warning of multiple parameters, easily leading to missed or false alarms.

[0005] Third, existing technologies are poorly adaptable to complex alignments and lack sufficient intelligence. Most existing monitoring methods are designed for straight-line jacking conditions. For curved bridge jacking, there is a lack of mature and efficient engineering solutions for easily and accurately characterizing and monitoring centerline deviation. At the same time, monitoring systems generally lack the ability to autonomously track moving targets. During the jacking process, the deflection monitoring points at the front end of the guide beam will move forward with the beam. Traditional fixed visual monitoring systems are easily lost when the target moves out of the field of view, requiring frequent manual intervention and resetting, and cannot achieve truly unattended automated monitoring.

[0006] To address the aforementioned issues, relevant technical solutions have been proposed. For example, Chinese Patent CN110634137A discloses a bridge deformation monitoring method based on machine vision, but it relies on image edge recognition technology, its accuracy is greatly affected by ambient lighting, and it is difficult to guarantee sub-millimeter level measurement accuracy at long distances and large fields of view; Chinese Patent CN117591984A discloses a bridge jacking attitude monitoring system based on GPS and a ranging sensor, but its GPS positioning accuracy is limited, and the ranging sensor needs to be installed on the moving beam, resulting in complex power supply and communication line layouts that are easily damaged by construction interference.

[0007] Therefore, the current field of bridge jacking construction monitoring urgently needs to develop an intelligent monitoring system and method that integrates high precision, automation, high safety, strong alignment adaptability, and multi-parameter coordination to solve the multiple technical problems of efficiency, safety, accuracy, and adaptability in existing technologies. Summary of the Invention

[0008] This invention addresses the shortcomings of existing technologies by providing a multi-parameter monitoring system and method for bridge jacking construction based on vision and laser fusion. It belongs to the category of remote, non-contact intelligent monitoring technology for key construction parameters such as temporary steel support deformation, beam centerline deviation, and guide beam deflection during bridge jacking construction, specifically achieving the following objectives: 1. It solves the problems of low efficiency and high safety risks of traditional manual measurement, and realizes all-weather, non-contact, and automated monitoring of key parameters of bridge jacking construction, completely avoiding the safety hazards of personnel entering high-risk work areas.

[0009] 2. To address the issues of insufficient monitoring accuracy and poor data coordination of single sensors, the system utilizes AI video recognition and laser precision ranging fusion technology to achieve millimeter-level high-precision measurement and to complete the collaborative monitoring and comprehensive analysis of multiple parameters, including steel support deformation, central axis deviation, and guide beam deflection.

[0010] 3. To address the problem of weak adaptability of existing technologies to complex alignments, a unified monitoring scheme applicable to both straight and curved jacking conditions is provided, enabling accurate characterization and calculation of centerline deviation under different alignments.

[0011] 4. Solve the automation bottleneck of easy loss of mobile monitoring targets and the need for frequent manual intervention, and realize autonomous tracking and unattended continuous monitoring of guide beam deflection monitoring points.

[0012] To achieve the above objectives, the present invention adopts the following technical solution: A multi-parameter monitoring method for bridge jacking construction based on vision and laser fusion includes the following steps: S1: System initialization and coordinate system alignment: This includes the following steps: S11. Control point layout: At least 3 control point prisms shall be laid out on the stable concrete structure outside the construction influence zone. The control points shall be arranged in an equilateral or isosceles triangle, and all monitoring points shall be located within the coverage area of ​​the triangle formed by the control points, following the principle of controlling the short side with the long side and the best frontal angle; if there is no existing stable structure, a 40cm×40cm×1.2m concrete pier shall be poured as the foundation for the control points. S12. Equipment Setup and Baseline Acquisition: The AI ​​video monitoring measuring instrument will be set up in a safe location with a line of sight to all control points and monitoring areas, and fixed with a 2m aluminum alloy tower. If the ground is not hardened, a 1m×1m×50cm concrete foundation with a depth of 30cm will be poured. The coordinates of all control points in the design coordinate system provided by the construction unit will be obtained. S13. Coordinate system transformation and accuracy verification: Control the AI ​​video monitoring measuring instrument to measure the coordinates of each control point in sequence, use the Bursa seven-parameter transformation model or the least squares method to calculate and fix the transformation parameters between the equipment coordinate system and the design coordinate system; measure the coordinates of one independent check point, and if the three-dimensional coordinate deviation of the check point is ≤ ±2mm, the coordinate system is deemed to be aligned; otherwise, the control points are re-laid out and the above steps are repeated. S14. Reference calibration mechanism setting: The system automatically sets all control point coordinates to be checked every 30 minutes. If the control point deviation exceeds the preset threshold, the system automatically recalculates the conversion parameters and calibrates the reference of all monitoring points to eliminate reference drift caused by factors such as temperature deformation and equipment micro-motion.

[0013] S2: Monitoring point layout and initial value acquisition: This includes the following steps: S21. Layout of steel support deformation monitoring points: On the front of the columns and key cross braces of the steel support to be monitored, attach 5cm×5cm high reflective patches; if night construction is required, replace them with welded and fixed right-angle prisms; each steel support shall have no less than 2 monitoring points. S22. Layout of centerline deviation monitoring points under different working conditions: For the straight-line jacking case: At least one checkerboard target is set up at the front, middle and rear of the beam in the jacking direction, for a total of ≥3 non-collinear targets, for global inverse algorithm calculation; at the same time, one checkerboard target is set up near the centerline of the front and rear cross sections of the beam for direct deviation method calculation and result cross-checking. Curved jacking case: Set up a checkerboard target at each of the four corner points of the jacking beam segment; receive the theoretical design coordinates corresponding to each jacking stroke of the four corner points of the beam segment exported by the construction unit based on the BIM model, and store them in the platform database; S23. Deflection monitoring point layout of guide beam: Weld a large deflection monitoring prism at the upper edge or centroid of the front end of the guide beam; no less than 4 points shall be laid out on the straight section and no less than 2 points shall be laid out on the curved section. S24. Initial value acquisition and filing: Generate automated monitoring tasks through the platform and set the patrol route; control the AI ​​video monitoring measuring instrument to continuously and independently measure all monitoring points 3 times, take the average value of the stable values ​​as the initial three-dimensional coordinates of each monitoring point, store them in the platform and complete the filing.

[0014] S3: Automated Periodic Monitoring and AI Intelligent Locator: Specifically includes the following steps: S31. Adaptive monitoring frequency strategy: Steel support deformation and centerline deviation monitoring: frequency is 1 time / hour when the jacking is stationary; automatically increased to 1 time / 5 minutes during jacking operation; further increased to 1 time / 2 minutes when an early warning is issued; Guide beam deflection monitoring: The entire process is continuously acquired at an equivalent rate of 100 frames / second, and the deflection time history curve is output in real time; Panoramic Image Acquisition: A panoramic image of the monitoring area is automatically acquired and archived once a day at 12 noon for the purpose of visual traceability of the construction process; S32. Automatic measurement of routine monitoring points: For steel support deformation monitoring points and centerline deviation monitoring points, the AI ​​video monitoring measuring instrument strictly follows the preset cruise route, rotates to each point in sequence, and executes the automated process of AI target center recognition → laser aiming and distance measurement → three-dimensional coordinate calculation → data upload. S33. Intelligent Tracking of Moving Deflection Monitoring Points: For deflection monitoring points that move forward with the beam, the following autonomous tracking process is executed: First, measure the current coordinates of at least three non-collinear checkerboard targets on the same rigid body; Solve for the optimal rigid body transformation (i.e., rotation matrix R and translation vector T) from the initial point set to the current point set. Rigid body transformation is applied to the initial coordinates of the deflection monitoring target to obtain the basic estimated coordinates; combined with the current jacking speed and monitoring cycle, the target movement compensation is calculated to further narrow the estimated spatial range. The bridge deflection visual monitoring instrument is controlled within a narrowed range. Using the YOLO target detection algorithm combined with sub-pixel corner point extraction technology, it accurately locates the center of the deflection monitoring prism and completes high-frequency measurement.

[0015] S4: Multi-source data processing and bias analysis: This includes the following steps: S41. Steel support deformation calculation: Calculate the difference between the current coordinates and the initial coordinates of each monitoring point, and obtain the cumulative value of vertical displacement and horizontal displacement and the rate of change per unit time. S42. Calculation of axis deviation under different working conditions: Linear jacking operation: Simultaneously call the dual-mode solution module to output the calculation results of the global inverse algorithm and the direct deviation method respectively; when the deviation between the two results is ≤2mm, the maximum value is taken as the final centerline deviation; when the deviation is >2mm, the monitoring frequency is automatically encrypted and an abnormal prompt is triggered. Curved jacking condition: The corner point comparison module is called to compare the real-time measured coordinates of the four corner points with the theoretical design coordinates corresponding to the current jacking stroke point by point, and calculate the northward deviation ΔN and eastward deviation ΔE of each corner point; by synthesizing the deviation vectors of the four corner points, the overall posture and centerline deviation of the curved beam segment are comprehensively evaluated. S43. Guide Beam Deflection Calculation and Correction: Calculates the difference between the current vertical coordinates and the initial vertical coordinates of the deflection monitoring target; automatically deducts elevation changes caused by beam climbing or descending; correction methods can be switched. Average slope correction method: Calculate the average slope of elevation change based on the difference in design elevation before and after the beam is launched and the total launching distance, and obtain the correction value by combining it with the current launching distance; Step-by-step theoretical correction method: Import the theoretical elevation data of the beam body provided by the construction unit for each step of the push-up stroke, and directly use this data for correction; the final guide beam deflection calculation formula is: guide beam deflection = current vertical coordinate - initial vertical coordinate - beam body elevation correction value.

[0016] S5: Dual-control collaborative early warning and information feedback: This includes the following steps: S51. Warning Threshold Setting: The platform uniformly manages the control thresholds of all monitoring items. The yellow warning value is 70% of the control value, and the red warning value is 100% of the control value. Among them, the vertical and horizontal displacement of the steel support adopts dual control of cumulative value + 2mm / h change rate, while the centerline deviation and guide beam deflection adopt single control of cumulative value. S52. Tiered Early Warning and Response: Yellow Alert: When any monitoring item reaches the yellow alert threshold, the platform sends an SMS reminder to the project manager, and the interface highlights the alert point, value, and trend; it automatically increases the monitoring frequency of the corresponding monitoring item and strengthens remote video surveillance; Red Alert: When any monitoring item reaches the red alert threshold, the on-site audible and visual alarm will be triggered immediately; at the same time, alarm information and real-time image snapshots of the warning point will be pushed to all relevant parties such as the project manager, supervision unit, and construction unit; the platform will automatically generate a written early warning report containing monitoring data curves and deformation analysis within 2 hours. S53. Information Feedback and Data Management: The platform displays data curves, deformation cloud maps, and equipment operating status for each monitoring item in real time; it supports AI comparison between the original image and the current image of the warning point to intuitively display the deformation situation; it automatically generates daily, weekly, and periodic monitoring reports, and all monitoring data and reports are permanently archived and can be exported and viewed at any time.

[0017] The monitoring system used in the aforementioned multi-parameter monitoring method for bridge jacking construction based on vision and laser fusion includes: Front-end intelligent sensing unit: includes at least one set of AI video monitoring and measuring instrument and one set of bridge deflection visual monitoring instrument; the AI ​​video monitoring and measuring instrument integrates a high-precision gimbal, laser ranging module and AI video recognition module, used to automatically search, aim and measure the three-dimensional coordinates of preset cooperative targets; the bridge deflection visual monitoring instrument integrates an industrial-grade fixed-focus camera and a high-frequency image acquisition module, used to perform sub-pixel level displacement measurement of cooperative targets at the front end of the guide beam; Multi-type cooperative target units include: control point large prisms set on stable benchmark points, reflective patches or small prisms set on steel support deformation monitoring points, checkerboard targets set on beams, and deflection monitoring large prisms set on the front end of guide beams. The control point prism serves as an absolute reference, aligning the equipment coordinate system with the design coordinate system at the initial moment. The steel support monitoring reflective patch, made of 5cm x 5cm highly reflective material, is attached to key stress nodes of the steel support. At night or in poor lighting conditions, it is replaced with a welded and fixed right-angle prism to ensure all-weather measurement capability. The checkerboard pattern of the target facilitates sub-pixel corner recognition by AI algorithms, while the prism provides an efficient reflective surface for laser ranging. The combination of these two features greatly improves the target recognition rate and measurement accuracy at long distances and against complex backgrounds. The target is mounted on a specially designed universal joint base, allowing for flexible orientation adjustment to ensure it always faces the monitoring equipment. The deflection monitoring prism serves as a cooperative target for high-frequency visual tracking.

[0018] Automated monitoring, control and calculation platform: The platform receives measurement data from the front-end intelligent sensing unit and has built-in modules for automatic coordinate system alignment and periodic verification, dual-mode calculation of straight jacking centerline deviation, corner point comparison of curved jacking centerline deviation, deflection monitoring target position estimation and AI repositioning based on rigid body kinematics, guide beam deflection elevation correction, adaptive monitoring frequency adjustment, and multi-parameter collaborative early warning based on dual control of cumulative value and rate of change. The dual-mode solution module for the deviation of the centerline of the linear jacking push includes: The global inverse algorithm calculation unit, based on the rigid body transformation principle, calculates the optimal rotation and translation transformation matrix of the beam relative to its initial state at the current moment based on the real-time measured coordinates of at least three non-collinear checkerboard targets on the beam. This transformation matrix is ​​then applied to two pre-defined feature points on the beam's central axis to obtain their current actual coordinates. The vertical distances from each feature point to the theoretical central axis are calculated, and the maximum value is taken as the current deviation of the beam's central axis. This method is theoretically rigorous, unaffected by minor offsets in the jacking direction, and applicable to most straight-line jacking conditions. Direct deviation method calculation unit: At least one checkerboard target is set up on each side of different cross sections of the beam along the jacking direction, and the vertical distance from each target to the theoretical centerline is calculated. The distance calculated in real time is compared with the theoretical distance measured at the initial installation, and the change is the centerline deviation at the corresponding cross section. This method is simple to calculate and has a clear physical meaning. It is suitable for working conditions where the jacking direction is strictly parallel to the design axis. The dual-mode solution module outputs the calculation results of the two methods and performs cross-checking. When the deviation between the calculation results of the two methods exceeds 2mm, the system automatically encrypts the monitoring frequency and triggers an abnormality prompt. The curve jacking centerline deviation corner point comparison module addresses the issue of the high complexity of directly calculating the vertical distance when the centerline of the curve jacking is curved. It employs a discrete point pose comparison method: receiving the stroke-by-stroke theoretical design coordinates of the four corner points of the jacked beam segment exported by the construction unit based on the BIM model, and automatically updating the theoretical coordinates after each jacking stroke; real-time monitoring of the actual spatial coordinates of the checkerboard targets placed at the four corner points of the beam segment, calculating the northward deviation ΔN and eastward deviation ΔE for each corner point; and comprehensively evaluating the overall pose and centerline deviation status of the curved beam segment through the synthesis of the deviation vectors of the four corner points. This method transforms the complex problem of curved spatial deviation into a comparison of the planar coordinates of four discrete points. The algorithm is simple, efficient, and highly practical in engineering, providing intuitive guidance for on-site jack stroke adjustments.

[0019] The deflection monitoring target position estimation and AI relocalization module based on rigid body kinematics performs the following steps: During the initialization phase, the initial coordinates of the large prism P0 for deflection monitoring at the front end of the guide beam, as well as the initial coordinates of at least three non-collinear checkerboard targets (Q1, Q2, Q3...) located on the same beam or guide beam, are measured and recorded. Periodic monitoring phase: First, measure the updated coordinates (Q1', Q2', Q3'...) of the above chessboard target at the current moment; Rigid body transformation solution: Solve for the optimal rigid body transformation from the initial point set {Q} to the current point set {Q'} to obtain the rotation matrix R and the translation vector T; Dual position estimation: Applying rigid body transformation to the initial coordinates P0 of the deflection monitoring target, the basic predicted coordinates P are obtained. pre =R・P0+T; Combine the current pushing speed and monitoring cycle to calculate the compensation amount generated by the target movement, and further narrow the estimated spatial coordinate range; AI-powered intelligent targeting: The bridge deflection visual monitoring instrument is controlled to accurately identify and lock the center of the deflection monitoring prism within the aforementioned narrowed range using the YOLO target detection algorithm combined with sub-pixel corner extraction technology, thus completing high-precision coordinate measurement.

[0020] This module effectively solves the technical problem that the deflection monitoring points easily move out of the field of view as the beam moves forward during the jacking process, requiring frequent manual intervention to reset them, and realizes autonomous tracking of the moving target.

[0021] The guide beam deflection elevation correction module: Automatically deducts elevation changes caused by the overall climbing or falling of the beam during the jacking process, eliminating deflection measurement distortion; supports two correction modes, which can be switched according to site conditions: Average slope correction method: Calculate the average slope of the beam elevation change based on the design elevation difference before and after the beam is launched and the total launching distance, and obtain the corresponding elevation correction value by combining it with the current launching distance; Step-by-stroke theoretical correction method: Import the theoretical elevation data of the beam corresponding to each launching stroke provided by the construction unit, and directly use this data for deflection correction. The final guide beam deflection calculation formula is: Guide beam deflection = Current vertical coordinate - Initial vertical coordinate - Beam elevation correction value.

[0022] The adaptive monitoring frequency adjustment module: The monitoring frequency of each monitoring item is automatically adjusted according to the jacking construction status to balance monitoring accuracy and equipment operating efficiency. Steel support deformation and centerline deviation monitoring: frequency is 1 time / hour when the jacking is stationary; automatically increased to 1 time / 5 minutes during jacking operation; further increased to 1 time / 2 minutes when an early warning is issued; Guide beam deflection monitoring: The entire process is continuously acquired at an equivalent rate of 100 frames / second, and the deflection time history curve is output in real time; A panoramic image of the monitoring area is automatically collected and archived every day at 12 noon for visual traceability of the construction process.

[0023] The multi-parameter collaborative early warning module based on dual-control indicators: Establish a tiered early warning mechanism that combines cumulative value and rate of change control, and uniformly manage the control thresholds for all monitored items: Warning level classification: Yellow warning value is 70% of the control value, and red warning value is 100% of the control value; among them, the vertical and horizontal displacement of the steel support is controlled by both cumulative value and change rate of 2mm / h, while the deviation of the central axis and the deflection of the guide beam are controlled by single cumulative value. Tiered early warning and response: 1. Yellow Alert: The platform sends SMS reminders to project managers, highlights the alert points and values ​​on the interface, automatically increases the monitoring frequency of the corresponding monitoring items, and strengthens remote video surveillance; 2. Red Alert: Immediately triggers the on-site audible and visual alarm, and simultaneously pushes alarm information and real-time image snapshots of the warning points to all relevant parties such as the project manager, supervisor, and construction party, and automatically generates a written warning bulletin within 2 hours; It supports AI comparison between the original image and the current image of the warning point, intuitively displaying the deformation situation and providing a visual basis for on-site handling.

[0024] The AI ​​video surveillance measuring instrument has the following characteristics: The high-precision gimbal has a horizontal motion range of 0° to 360° and a vertical motion range of -90° to +90°. The camera has an image resolution of 1920×1080px and features optical zoom of more than 32x. Automatic monitoring accuracy: ±0.5mm (measuring range ≤50m), ±1mm (measuring range 50m~100m); Supports remote visual deployment of monitoring points, automatic cruise measurement, automatic target tracking and aiming; automatically checks the coordinates of the control point prism every 30 minutes and calibrates the measurement benchmark. The bridge deflection visual monitoring instrument has the following characteristics: It uses an industrial-grade fixed-focus lens with an image resolution of 4096×2460px; Measurement accuracy: 0.1mm / 10m, 0.2mm / 20m, 0.3mm / 30m; The physical frame rate is 25 frames per second, and displacement monitoring is achieved at an equivalent of 100 frames per second through frame interpolation technology. It supports simultaneous real-time monitoring of up to 7 monitoring points.

[0025] The checkerboard target is a 240mm×240mm square black and white checkerboard, with prisms integrated on its back or side; the target is mounted on the beam via a special support and universal joint, allowing for adjustable spatial angle.

[0026] Compared with existing technologies, this invention achieves the following beneficial effects through the fusion of AI vision and laser ranging technology, differentiated linear monitoring algorithms, and a fully automated process mechanism: 1. Enables non-contact remote monitoring, allowing for continuous operation around the clock: All measurements are performed by automated equipment located in safe areas, avoiding the need for surveyors to enter high-risk work environments such as under the jacking beam or in the support area.

[0027] 2. Achieve fully automated monitoring throughout the entire process to improve monitoring efficiency: The system automates the entire process from target identification and coordinate measurement to data processing and early warning feedback. A single device can manage dozens of monitoring points simultaneously, with each point measurement taking tens of seconds, improving efficiency by more than 10 times compared to traditional manual point-by-point measurement. It reduces process interruptions caused by monitoring wait times and automatically generates monitoring data and reports.

[0028] 3. Achieve millimeter-level high-precision measurement and improve data reliability: By integrating AI video recognition and laser precision ranging technology, the steel support deformation monitoring accuracy reaches ±0.5mm@50m and ±1mm@100m, the centerline deviation monitoring accuracy reaches ±1mm / 100m, and the guide beam deflection monitoring accuracy reaches 0.1mm@10m. The linear jacking process employs dual-mode calculation results cross-checking, with the coordinate system automatically recalibrated every 30 minutes. Guide beam deflection is corrected for beam elevation changes, reducing measurement errors caused by factors such as reference drift and beam lifting / lowering. Guide beam deflection uses equivalent 100Hz high-frequency sampling to capture the dynamic response during the jacking process.

[0029] 4. Simultaneously supports both linear and curved jacking operations: A single system can adapt to two different track shapes without changing equipment or algorithms. Straight-line jacking employs a dual-mode solution combining a global inverse algorithm and a direct deviation method; curved jacking uses a four-corner point pose comparison method based on BIM-based stroke-by-stroke theoretical coordinates, transforming the spatial deviation problem of curves into discrete point coordinate comparison. It is suitable for jacking construction of steel-concrete composite beams and steel box girders for various types of bridges, including overpasses, river-crossing bridges, and interchanges.

[0030] 5. Achieve autonomous tracking and adaptive monitoring of moving targets: By combining rigid body transformation with jacking speed prediction, the system solves the problem of guide beam deflection monitoring points being easily lost as the beam moves forward, requiring frequent manual intervention and resetting. The system supports adaptive monitoring frequency adjustment, which can adjust the monitoring frequency according to the jacking construction status; it establishes a hierarchical early warning mechanism based on dual control of cumulative value and rate of change, which can automatically identify anomalies and trigger corresponding level of early warning.

[0031] 6. Achieve digital management of monitoring data and reduce labor costs: All monitoring data is digitally stored and managed, allowing for the review of historical data, generation of data curves and deformation cloud maps, providing traceable evidence for construction quality assessment, safety evaluation, and subsequent operation and maintenance. Compared with traditional manual monitoring, this reduces the number of on-site surveyors by more than 70%. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the deviation measurement of the centerline of the curved segment jacking in this invention; Figure 2 This is a schematic diagram of the deflection monitoring of the guide beam during the curved segment jacking of the present invention; Figure 3 This is a schematic diagram of the steel support monitoring during the curved segment jacking of the present invention; Figure 4 This is a schematic diagram of the deviation measurement of the centerline of the straight segment jacking according to the present invention; Figure 5 This is a schematic diagram of the deflection monitoring of the guide beam during the straight-segment jacking of the present invention; Figure 6 This is a schematic diagram of the steel support monitoring during the straight-segment jacking of the present invention; Figure 7 This is a schematic diagram of the multifunctional checkerboard target of the present invention; In the diagram: 1-AI video monitoring and measuring instrument; 2-bridge deflection visual monitoring instrument; 3-monitoring point; 4-control point; 5-multifunctional checkerboard target. Detailed Implementation

[0033] The present invention will be further described in detail below with reference to specific engineering embodiments and accompanying drawings. The described embodiments are only a part of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the technical solutions of the present invention without creative effort are within the scope of protection of the present invention.

[0034] The overall implementation process of the AI-based visual fusion-based remote intelligent monitoring system for bridge jacking construction is as follows: First, the system hardware deployment and coordinate system alignment are completed to establish a unified measurement benchmark; then, monitoring targets are set up according to the differences in the jacking alignment, and initial benchmark data is collected; the automated monitoring task is started, and the system adaptively adjusts the monitoring frequency according to the construction status, performs automatic cruise measurement on conventional monitoring points, and autonomously tracks moving deflection monitoring points; parameters such as steel support deformation, centerline deviation, and guide beam deflection are calculated in real time; when the monitoring data reaches the warning threshold, a graded warning is triggered and information is pushed.

[0035] Example 1

[0036] A multi-parameter monitoring method for bridge jacking construction based on vision and laser fusion includes the following steps: S1: System initialization and coordinate system alignment: This includes the following steps: S11. Control point layout: At least 3 control point prisms shall be laid out on the stable concrete structure outside the construction influence zone. The control points shall be arranged in an equilateral or isosceles triangle, and all monitoring points shall be located within the coverage area of ​​the triangle formed by the control points, following the principle of controlling the short side with the long side and the best frontal angle; if there is no existing stable structure, a 40cm×40cm×1.2m concrete pier shall be poured as the foundation for the control points. S12. Equipment Setup and Baseline Acquisition: The AI ​​video monitoring measuring instrument will be set up in a safe location with a line of sight to all control points and monitoring areas, and fixed with a 2m aluminum alloy tower. If the ground is not hardened, a 1m×1m×50cm concrete foundation with a depth of 30cm will be poured. The coordinates of all control points in the design coordinate system provided by the construction unit will be obtained. S13. Coordinate system transformation and accuracy verification: Control the AI ​​video monitoring measuring instrument to measure the coordinates of each control point in sequence, use the Bursa seven-parameter transformation model or the least squares method to calculate and fix the transformation parameters between the equipment coordinate system and the design coordinate system; measure the coordinates of one independent check point, and if the three-dimensional coordinate deviation of the check point is ≤ ±2mm, the coordinate system is deemed to be aligned; otherwise, the control points are re-laid out and the above steps are repeated. S14. Reference calibration mechanism setting: The system automatically sets all control point coordinates to be checked every 30 minutes. If the control point deviation exceeds the preset threshold, the system automatically recalculates the conversion parameters and calibrates the reference of all monitoring points to eliminate reference drift caused by factors such as temperature deformation and equipment micro-motion.

[0037] S2: Monitoring point layout and initial value acquisition: This includes the following steps: S21. Layout of steel support deformation monitoring points: On the front of the columns and key cross braces of the steel support to be monitored, attach 5cm×5cm high reflective patches; if night construction is required, replace them with welded and fixed right-angle prisms; each steel support shall have no less than 2 monitoring points. S22. Layout of centerline deviation monitoring points under different working conditions: For the straight-line jacking case: At least one checkerboard target is set up at the front, middle and rear of the beam in the jacking direction, for a total of ≥3 non-collinear targets, for global inverse algorithm calculation; at the same time, one checkerboard target is set up near the centerline of the front and rear cross sections of the beam for direct deviation method calculation and result cross-checking. Curved jacking case: Set up a checkerboard target at each of the four corner points of the jacking beam segment; receive the theoretical design coordinates corresponding to each jacking stroke of the four corner points of the beam segment exported by the construction unit based on the BIM model, and store them in the platform database; S23. Deflection monitoring point layout of guide beam: Weld a large deflection monitoring prism at the upper edge or centroid of the front end of the guide beam; no less than 4 points shall be laid out on the straight section and no less than 2 points shall be laid out on the curved section. S24. Initial value acquisition and filing: Generate automated monitoring tasks through the platform and set the patrol route; control the AI ​​video monitoring measuring instrument to continuously and independently measure all monitoring points 3 times, take the average value of the stable values ​​as the initial three-dimensional coordinates of each monitoring point, store them in the platform and complete the filing.

[0038] S3: Automated Periodic Monitoring and AI Intelligent Locator: Specifically includes the following steps: S31. Adaptive monitoring frequency strategy: Steel support deformation and centerline deviation monitoring: frequency is 1 time / hour when the jacking is stationary; automatically increased to 1 time / 5 minutes during jacking operation; further increased to 1 time / 2 minutes when an early warning is issued; Guide beam deflection monitoring: The entire process is continuously acquired at an equivalent rate of 100 frames / second, and the deflection time history curve is output in real time; Panoramic Image Acquisition: A panoramic image of the monitoring area is automatically acquired and archived once a day at 12 noon for the purpose of visual traceability of the construction process; S32. Automatic measurement of routine monitoring points: For steel support deformation monitoring points and centerline deviation monitoring points, the AI ​​video monitoring measuring instrument strictly follows the preset cruise route, rotates to each point in sequence, and executes the automated process of AI target center recognition → laser aiming and distance measurement → three-dimensional coordinate calculation → data upload. S33. Intelligent Tracking of Moving Deflection Monitoring Points: For deflection monitoring points that move forward with the beam, the following autonomous tracking process is executed: First, measure the current coordinates of at least three non-collinear checkerboard targets on the same rigid body; Solve for the optimal rigid body transformation (i.e., rotation matrix R and translation vector T) from the initial point set to the current point set. Rigid body transformation is applied to the initial coordinates of the deflection monitoring target to obtain the basic estimated coordinates; combined with the current jacking speed and monitoring cycle, the target movement compensation is calculated to further narrow the estimated spatial range. The bridge deflection visual monitoring instrument is controlled within a narrowed range. Using the YOLO target detection algorithm combined with sub-pixel corner point extraction technology, it accurately locates the center of the deflection monitoring prism and completes high-frequency measurement.

[0039] S4: Multi-source data processing and bias analysis: This includes the following steps: S41. Steel support deformation calculation: Calculate the difference between the current coordinates and the initial coordinates of each monitoring point, and obtain the cumulative value of vertical displacement and horizontal displacement and the rate of change per unit time. S42. Calculation of axis deviation under different working conditions: Linear jacking operation: Simultaneously call the dual-mode solution module to output the calculation results of the global inverse algorithm and the direct deviation method respectively; when the deviation between the two results is ≤2mm, the maximum value is taken as the final centerline deviation; when the deviation is >2mm, the monitoring frequency is automatically encrypted and an abnormal prompt is triggered. Curved jacking condition: The corner point comparison module is called to compare the real-time measured coordinates of the four corner points with the theoretical design coordinates corresponding to the current jacking stroke point by point, and calculate the northward deviation ΔN and eastward deviation ΔE of each corner point; by synthesizing the deviation vectors of the four corner points, the overall posture and centerline deviation of the curved beam segment are comprehensively evaluated. S43. Guide Beam Deflection Calculation and Correction: Calculates the difference between the current vertical coordinates and the initial vertical coordinates of the deflection monitoring target; automatically deducts elevation changes caused by beam climbing or descending; correction methods can be switched. Average slope correction method: Calculate the average slope of elevation change based on the difference in design elevation before and after the beam is launched and the total launching distance, and obtain the correction value by combining it with the current launching distance; Step-by-step theoretical correction method: Import the theoretical elevation data of the beam body provided by the construction unit for each step of the push-up stroke, and directly use this data for correction; the final guide beam deflection calculation formula is: guide beam deflection = current vertical coordinate - initial vertical coordinate - beam body elevation correction value.

[0040] S5: Dual-control collaborative early warning and information feedback: This includes the following steps: S51. Warning Threshold Setting: The platform uniformly manages the control thresholds of all monitoring items. The yellow warning value is 70% of the control value, and the red warning value is 100% of the control value. Among them, the vertical and horizontal displacement of the steel support adopts dual control of cumulative value + 2mm / h change rate, while the centerline deviation and guide beam deflection adopt single control of cumulative value. S52. Tiered Early Warning and Response: Yellow Alert: When any monitoring item reaches the yellow alert threshold, the platform sends an SMS reminder to the project manager, and the interface highlights the alert point, value, and trend; it automatically increases the monitoring frequency of the corresponding monitoring item and strengthens remote video surveillance; Red Alert: When any monitoring item reaches the red alert threshold, the on-site audible and visual alarm will be triggered immediately; at the same time, alarm information and real-time image snapshots of the warning point will be pushed to all relevant parties such as the project manager, supervision unit, and construction unit; the platform will automatically generate a written early warning report containing monitoring data curves and deformation analysis within 2 hours. S53. Information Feedback and Data Management: The platform displays data curves, deformation cloud maps, and equipment operating status for each monitoring item in real time; it supports AI comparison between the original image and the current image of the warning point to intuitively display the deformation situation; it automatically generates daily, weekly, and periodic monitoring reports, and all monitoring data and reports are permanently archived and can be exported and viewed at any time.

[0041] Example 2

[0042] The monitoring system used in the aforementioned multi-parameter monitoring method for bridge jacking construction based on vision and laser fusion includes: Front-end intelligent sensing unit: includes at least one set of AI video monitoring and measuring instrument and one set of bridge deflection visual monitoring instrument; the AI ​​video monitoring and measuring instrument integrates a high-precision gimbal, laser ranging module and AI video recognition module, used to automatically search, aim and measure the three-dimensional coordinates of preset cooperative targets; the bridge deflection visual monitoring instrument integrates an industrial-grade fixed-focus camera and a high-frequency image acquisition module, used to perform sub-pixel level displacement measurement of cooperative targets at the front end of the guide beam; Multi-type cooperative target units include: control point large prisms set on stable benchmark points, reflective patches or small prisms set on steel support deformation monitoring points, checkerboard targets set on beams, and deflection monitoring large prisms set on the front end of guide beams. The control point prism serves as an absolute reference, aligning the equipment coordinate system with the design coordinate system at the initial moment. The steel support monitoring reflective patch, made of 5cm x 5cm highly reflective material, is attached to key stress nodes of the steel support. At night or in poor lighting conditions, it is replaced with a welded and fixed right-angle prism to ensure all-weather measurement capability. The checkerboard pattern of the target facilitates sub-pixel corner recognition by AI algorithms, while the prism provides an efficient reflective surface for laser ranging. The combination of these two features greatly improves the target recognition rate and measurement accuracy at long distances and against complex backgrounds. The target is mounted on a specially designed universal joint base, allowing for flexible orientation adjustment to ensure it always faces the monitoring equipment. The deflection monitoring prism serves as a cooperative target for high-frequency visual tracking.

[0043] Automated monitoring, control and calculation platform: The platform receives measurement data from the front-end intelligent sensing unit and has built-in modules for automatic coordinate system alignment and periodic verification, dual-mode calculation of straight jacking centerline deviation, corner point comparison of curved jacking centerline deviation, deflection monitoring target position estimation and AI repositioning based on rigid body kinematics, guide beam deflection elevation correction, adaptive monitoring frequency adjustment, and multi-parameter collaborative early warning based on dual control of cumulative value and rate of change. The dual-mode solution module for the deviation of the centerline of the linear jacking push includes: The global inverse algorithm calculation unit, based on the rigid body transformation principle, calculates the optimal rotation and translation transformation matrix of the beam relative to its initial state at the current moment based on the real-time measured coordinates of at least three non-collinear checkerboard targets on the beam. This transformation matrix is ​​then applied to two pre-defined feature points on the beam's central axis to obtain their current actual coordinates. The vertical distances from each feature point to the theoretical central axis are calculated, and the maximum value is taken as the current deviation of the beam's central axis. This method is theoretically rigorous, unaffected by minor offsets in the jacking direction, and applicable to most straight-line jacking conditions. Direct deviation method calculation unit: At least one checkerboard target is set up on each side of different cross sections of the beam along the jacking direction, and the vertical distance from each target to the theoretical centerline is calculated. The distance calculated in real time is compared with the theoretical distance measured at the initial installation, and the change is the centerline deviation at the corresponding cross section. This method is simple to calculate and has a clear physical meaning. It is suitable for working conditions where the jacking direction is strictly parallel to the design axis. The dual-mode solution module outputs the calculation results of the two methods and performs cross-checking. When the deviation between the calculation results of the two methods exceeds 2mm, the system automatically encrypts the monitoring frequency and triggers an abnormality prompt. The curve jacking centerline deviation corner point comparison module addresses the issue of the high complexity of directly calculating the vertical distance when the centerline of the curve jacking is curved. It employs a discrete point pose comparison method: receiving the stroke-by-stroke theoretical design coordinates of the four corner points of the jacked beam segment exported by the construction unit based on the BIM model, and automatically updating the theoretical coordinates after each jacking stroke; real-time monitoring of the actual spatial coordinates of the checkerboard targets placed at the four corner points of the beam segment, calculating the northward deviation ΔN and eastward deviation ΔE for each corner point; and comprehensively evaluating the overall pose and centerline deviation status of the curved beam segment through the synthesis of the deviation vectors of the four corner points. This method transforms the complex problem of curved spatial deviation into a comparison of the planar coordinates of four discrete points. The algorithm is simple, efficient, and highly practical in engineering, providing intuitive guidance for on-site jack stroke adjustments.

[0044] The deflection monitoring target position estimation and AI relocalization module based on rigid body kinematics performs the following steps: During the initialization phase, the initial coordinates of the large prism P0 for deflection monitoring at the front end of the guide beam, as well as the initial coordinates of at least three non-collinear checkerboard targets (Q1, Q2, Q3...) located on the same beam or guide beam, are measured and recorded. Periodic monitoring phase: First, measure the updated coordinates (Q1', Q2', Q3'...) of the above chessboard target at the current moment; Rigid body transformation solution: Solve for the optimal rigid body transformation from the initial point set {Q} to the current point set {Q'} to obtain the rotation matrix R and the translation vector T; Dual position estimation: Applying rigid body transformation to the initial coordinates P0 of the deflection monitoring target, the basic predicted coordinates P are obtained.pre =R・P0+T; Combine the current pushing speed and monitoring cycle to calculate the compensation amount generated by the target movement, and further narrow the estimated spatial coordinate range; AI-powered intelligent targeting: The bridge deflection visual monitoring instrument is controlled to accurately identify and lock the center of the deflection monitoring prism within the aforementioned narrowed range using the YOLO target detection algorithm combined with sub-pixel corner extraction technology, thus completing high-precision coordinate measurement.

[0045] This module effectively solves the technical problem that the deflection monitoring points easily move out of the field of view as the beam moves forward during the jacking process, requiring frequent manual intervention to reset them, and realizes autonomous tracking of the moving target.

[0046] The guide beam deflection elevation correction module: Automatically deducts elevation changes caused by the overall climbing or falling of the beam during the jacking process, eliminating deflection measurement distortion; supports two correction modes, which can be switched according to site conditions: Average slope correction method: Calculate the average slope of the beam elevation change based on the design elevation difference before and after the beam is launched and the total launching distance, and obtain the corresponding elevation correction value by combining it with the current launching distance; Step-by-stroke theoretical correction method: Import the theoretical elevation data of the beam corresponding to each launching stroke provided by the construction unit, and directly use this data for deflection correction. The final guide beam deflection calculation formula is: Guide beam deflection = Current vertical coordinate - Initial vertical coordinate - Beam elevation correction value.

[0047] The adaptive monitoring frequency adjustment module: The monitoring frequency of each monitoring item is automatically adjusted according to the jacking construction status to balance monitoring accuracy and equipment operating efficiency. Steel support deformation and centerline deviation monitoring: frequency is 1 time / hour when the jacking is stationary; automatically increased to 1 time / 5 minutes during jacking operation; further increased to 1 time / 2 minutes when an early warning is issued; Guide beam deflection monitoring: The entire process is continuously acquired at an equivalent rate of 100 frames / second, and the deflection time history curve is output in real time; A panoramic image of the monitoring area is automatically collected and archived every day at 12 noon for visual traceability of the construction process.

[0048] The multi-parameter collaborative early warning module based on dual-control indicators: Establish a tiered early warning mechanism that combines cumulative value and rate of change control, and uniformly manage the control thresholds for all monitored items: Warning level classification: Yellow warning value is 70% of the control value, and red warning value is 100% of the control value; among them, the vertical and horizontal displacement of the steel support is controlled by both cumulative value and change rate of 2mm / h, while the deviation of the central axis and the deflection of the guide beam are controlled by single cumulative value. Tiered early warning and response: 1. Yellow Alert: The platform sends SMS reminders to project managers, highlights the alert points and values ​​on the interface, automatically increases the monitoring frequency of the corresponding monitoring items, and strengthens remote video surveillance; 2. Red Alert: Immediately triggers the on-site audible and visual alarm, and simultaneously pushes alarm information and real-time image snapshots of the warning points to all relevant parties such as the project manager, supervisor, and construction party, and automatically generates a written warning bulletin within 2 hours; It supports AI comparison between the original image and the current image of the warning point, intuitively displaying the deformation situation and providing a visual basis for on-site handling.

[0049] The AI ​​video surveillance measuring instrument has the following characteristics: The high-precision gimbal has a horizontal motion range of 0° to 360° and a vertical motion range of -90° to +90°. The camera has an image resolution of 1920×1080px and features optical zoom of more than 32x. Automatic monitoring accuracy: ±0.5mm (measuring range ≤50m), ±1mm (measuring range 50m~100m); Supports remote visual deployment of monitoring points, automatic cruise measurement, automatic target tracking and aiming; automatically checks the coordinates of the control point prism every 30 minutes and calibrates the measurement benchmark. The bridge deflection visual monitoring instrument has the following characteristics: It uses an industrial-grade fixed-focus lens with an image resolution of 4096×2460px; Measurement accuracy: 0.1mm / 10m, 0.2mm / 20m, 0.3mm / 30m; The physical frame rate is 25 frames per second, and displacement monitoring is achieved at an equivalent of 100 frames per second through frame interpolation technology. It supports simultaneous real-time monitoring of up to 7 monitoring points.

[0050] like Figure 7 As shown, the checkerboard target is a 240mm×240mm square black and white checkerboard, with prisms integrated on its back or side; the target is mounted on the beam via a special support and universal joint, allowing for adjustable spatial angle.

[0051] Example 3

[0052] This embodiment uses the Qin-Shen Expressway Raoyanghe Grand Bridge steel-concrete composite beam jacking construction project as an example to illustrate the monitoring system and method described in the above embodiment.

[0053] The bridge is a multi-span straight bridge with a maximum jacking distance of 120m per span. The self-weight of the beam is about 2500t. The design requires the deviation of the centerline to be controlled within ±10mm and the deflection of the guide beam to be controlled within 130mm.

[0054] 1. System Deployment (1) Deployment of monitoring equipment: In a safe area approximately 80m from the farthest monitoring point on the side of the bridge site, an AI video monitoring and measuring instrument (HMS500 model) was installed, fixed to a 2m high aluminum alloy tower. The tower foundation was a 1m×1m×50cm concrete structure (buried 30cm deep). This equipment is used to monitor the displacement of the steel support and the deviation of the centerline.

[0055] At the top of pier #0 behind the cap beam, a bridge deflection visual monitoring instrument (HMS100-LD type) is installed and directly fixed to the top surface of the cap beam with expansion bolts. The coordinates of the center point of the instrument are laid out in advance by the construction unit to ensure the horizontal stability of the instrument. This instrument is aligned with the front end of the guide beam for high-frequency monitoring of the guide beam deflection.

[0056] An outdoor audible and visual alarm module is installed on site in a location easily observed by construction personnel and linked with the automated monitoring, control, and calculation platform.

[0057] (2) Cooperative target deployment: Control points: Four control point prisms are set up on the piers of No. 0 and No. 5, which are not affected by construction, in an isosceles triangle arrangement. All monitoring points are located within the triangle coverage area formed by the control points.

[0058] like Figure 6 As shown, the steel support monitoring points are as follows: two 5cm×5cm high reflective patches are pasted on the front of each of the five key load-bearing steel support columns. In this embodiment, no nighttime construction is carried out, and small prisms are not used.

[0059] like Figure 4 As shown, the monitoring points along the central axis are as follows: along the jacking direction, three 240mm×240mm checkerboard fusion targets are installed at the side webs of the beam at distances of 10m, 60m, and 110m from the jacking starting point for global inverse algorithm calculation; one checkerboard fusion target is installed near the central axis of each of the two cross sections at the front and rear ends of the beam for direct deviation method calculation and result cross-checking.

[0060] like Figure 5 As shown, deflection monitoring points: Four large prisms are welded at the centroid position of the front end of the guide beam for monitoring the deflection of the guide beam.

[0061] (3) Platform deployment: The automated monitoring, control, and calculation platform is deployed on the on-site industrial computer and communicates wirelessly with the front-end devices via an IoT card (60G / month) to achieve data transmission and remote control.

[0062] 2. Implementation Steps (1) System initialization and coordinate system alignment: Obtain the design coordinates of four control points provided by the construction unit. Control an AI video monitoring measuring instrument to sequentially measure the coordinates of each control point, and use the Bursa-Taylor seven-parameter transformation model to calculate and solidify the transformation parameters between the equipment coordinate system and the design coordinate system. Measure the coordinates of one independent checkpoint, verifying a deviation of 1.2mm, meeting the accuracy requirement of ≤±2mm, thus completing coordinate system alignment. The system is set to automatically verify the control point coordinates every 30 minutes.

[0063] (2) Initial value acquisition: Create a straight-line segment initial value acquisition task through the platform and set the cruise route. Control the AI ​​video monitoring measuring instrument to continuously and independently measure all monitoring points three times, take the average of the stable values ​​as the initial three-dimensional coordinates of each monitoring point, and store them in the platform for record-keeping.

[0064] (3) Monitoring parameter settings: Warning thresholds: The cumulative warning value for vertical / horizontal displacement of the steel support is 1 / 500 of the support height, and the warning value for the rate of change is 2mm / h; the yellow warning value for centerline deviation is 7mm, and the red warning value is 10mm; the yellow warning value for guide beam deflection is 110mm, and the red warning value is 130mm.

[0065] Monitoring frequency: The deviation between the steel support and the central axis is monitored once per hour when the jacking is stationary, and automatically increased to once every 5 minutes when the jacking operation is in progress; the guide beam deflection is continuously collected at an equivalent rate of 100 frames per second throughout the entire process; panoramic images are automatically collected and archived at 12 noon every day.

[0066] (4) Automated monitoring and data processing: The system starts up automatically: The AI ​​video surveillance measuring instrument sequentially measures the coordinates of the steel support and the central axis target along a preset cruise route. The platform simultaneously calls the dual-mode calculation module to calculate the central axis deviation. When the deviation between the two methods is ≤2mm, the maximum value is taken as the final deviation.

[0067] For deflection monitoring points, the system first acquires the current coordinates of three checkerboard targets on the same rigid body before each measurement, calculates the rigid body transformation matrix, and estimates the position of the deflection targets based on the current jacking speed (approximately 3 m / h), guiding the bridge deflection visual monitoring instrument to perform high-frequency measurements. The platform automatically uses the average slope correction method to deduct the beam elevation change and calculates the guide beam deflection value.

[0068] (5) Early warning and response: At a certain point during the jacking process, the platform, using a global inverse algorithm, determined the centerline deviation to be 8.5mm. The direct deviation method yielded a front-end cross-section deviation of 9.0mm and a rear-end cross-section deviation of 4mm. The combined deviation of the two methods was 0.5mm, and the maximum value of 9.0mm was taken, reaching the yellow warning threshold. The platform sent an SMS alert to the project management personnel, highlighted the warning point on the interface, and increased the monitoring frequency to once every two minutes. Based on this, the construction team fine-tuned the stroke of the left-side jacking jack. After two monitoring cycles, the deviation decreased to 6mm, and the warning was lifted.

[0069] As the guide beam approached the top of the pier, the deflection monitoring instrument measured a maximum deflection of 125mm, reaching the red warning threshold. The platform immediately triggered an on-site audible and visual alarm, and simultaneously pushed alarm information and real-time images to all relevant parties. On-site personnel quickly slowed down the jacking speed and reinforced the temporary support on the top of the pier. After the guide beam safely passed the pier, the deflection dropped to 90mm, and the warning was lifted.

[0070] Implementation effect

[0071] In this embodiment, the entire jacking construction cycle took approximately 20 days. The system operated continuously and stably, completing over 6,000 effective monitoring points with a data integrity rate of 99.5%. The deviation of the centerline was consistently controlled within the design requirement of ±10mm, and the maximum deflection of the guide beam did not exceed the control value of 130mm. Compared with traditional manual monitoring, this reduced the number of on-site surveyors by 3, increased monitoring efficiency by more than 10 times, and prevented any safety accidents or project delays caused by monitoring.

[0072] Example 4

[0073] This embodiment illustrates the monitoring system and method described in the above embodiments using the jacking project of the curved bridge on ramp D of the Liuhegou Interchange. The bridge is located on a circular curve with a radius of 400m, and the jacking alignment is complex. The design requires that the deviation of the plane position of the corner points of the beam segments be controlled within ±30mm.

[0074] 1. System Deployment and Adjustment: This embodiment makes adaptive adjustments to the straight-segment system for curved jacking conditions: like Figure 3 As shown, the monitoring equipment deployment is consistent with that of the straight section, still using one AI video monitoring and measuring instrument and one bridge deflection visual monitoring instrument.

[0075] like Figure 1 As shown, the monitoring points for the central axis have been adjusted: the direct monitoring method of the central axis is no longer used, but instead the corner point comparison method of the beam segment is adopted. A checkerboard fusion target is welded to each of the four corner points of the jacked beam segment: the left side of the front end, the right side of the front end, the left side of the rear end, and the right side of the rear end.

[0076] Theoretical coordinate import: Receive the theoretical design coordinates corresponding to each jacking stroke of the four corner points of the beam segment exported by the construction unit based on the BIM model, store them in the platform database, and automatically update the theoretical coordinates after each jacking stroke is completed.

[0077] like Figure 2 As shown, the deflection monitoring points are adjusted by welding a large prism to the front end of each of the two guide beams, resulting in a total of two deflection monitoring points.

[0078] 2. Implementation steps: (1) System initialization and coordinate system alignment: Similar to Example 1, control points were set up, equipment was installed, and the coordinate system was aligned. The deviation of the independent checkpoints was 1.5mm, which met the accuracy requirements.

[0079] (2) Initial value acquisition: The initial coordinates of the four corner checkerboard targets and the two deflection monitoring prisms were collected, and the average value was taken after three consecutive measurements and stored in the platform.

[0080] (3) Monitoring parameter settings: Warning thresholds: Yellow warning value for the plane position deviation of the corner point of the beam segment is 21mm (70% control value), and red warning value is 30mm; other parameters are the same as those for the straight segment.

[0081] (4) Automated monitoring and data processing: The system measures the actual coordinates of the four corner targets every 5 minutes, calls the curve jacking centerline deviation corner point comparison module, compares the actual coordinates with the theoretical coordinates corresponding to the current jacking stroke point by point, calculates the northward deviation ΔN and eastward deviation ΔE of each corner point, and outputs the deviation vector map.

[0082] The guide beam deflection monitoring adopts the stroke-by-stroke theoretical correction method, which imports the theoretical elevation data of the beam for each jacking stroke provided by the construction unit to correct the deflection measurement results.

[0083] (5) Early warning and response: During the jacking process, the platform measured a deviation of +25mm (to the right) at the left corner of the front end of the beam and +20mm at the right corner, reaching the yellow warning threshold. The platform sent an SMS alert and displayed a deviation vector diagram of the four corner points. Based on the deviation vector diagram, the on-site commander adjusted the stroke of the left-side jacks to be greater than that of the right-side jacks, gradually correcting the rightward deviation of the beam. After three monitoring cycles, the deviations at each corner point returned to within 15mm, and the warning was lifted.

[0084] 3. Implementation Results: This embodiment successfully achieved real-time monitoring and deviation control of the beam's position during curved jacking, providing intuitive guidance for adjusting the jack stroke on-site through the deviation vectors at the four corner points. Throughout the jacking process, the maximum deviation at the beam segment corners was 25mm, which did not exceed the design control value of 30mm. This demonstrates that the present invention has good adaptability to curved jacking conditions and solves the problem of traditional techniques' difficulty in effectively monitoring the deviation of the centerline during curved jacking.

[0085] This invention is not limited to the above embodiments, for example: 1. When the monitoring distance exceeds 100m, the number of AI video monitoring measuring instruments can be increased, or equipment with higher magnification optical zoom can be used to ensure measurement accuracy.

[0086] 2. When nighttime monitoring of steel bracket deformation is required, the reflective patch can be replaced with a welded and fixed right-angle prism.

[0087] 3. The method for correcting the deflection elevation of the guide beam can be flexibly switched according to the site conditions. When the construction unit can provide accurate theoretical elevation for each stroke, the theoretical correction method for each stroke should be given priority.

[0088] 4. The warning threshold can be adjusted according to the design requirements of different projects, and is not limited to the values ​​in the above embodiments.

[0089] Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this invention shall be included within the scope of protection of this invention.

Claims

1. A multi-parameter monitoring method for bridge jacking construction based on vision and laser fusion, characterized in that, Includes the following steps: S1: System initialization and coordinate system alignment: Set up at least 3 control point prisms on stable benchmarks outside the construction influence area and obtain their design coordinate system coordinates; set up the AI ​​video monitoring measuring instrument in a position with good visibility, measure the coordinates of the control points, calculate the transformation parameters between the equipment coordinate system and the design coordinate system, and complete the alignment after verifying that the coordinate deviation of the independent checkpoints is ≤±2mm. S2: Monitoring point layout and initial value acquisition: According to the monitoring plan, the corresponding targets are laid out. The AI ​​video monitoring measuring instrument automatically cruises and continuously and independently measures all monitoring points three times. The average value of the stable values ​​is taken as the initial three-dimensional coordinates and stored in the platform. S3: Automated periodic monitoring and AI intelligent location: The platform starts monitoring tasks based on an adaptive monitoring frequency strategy; for steel support deformation and axis deviation monitoring points, the AI ​​video monitoring measuring instrument measures according to a preset route; for deflection monitoring points, the position estimation and AI repositioning process is executed to guide the bridge deflection visual monitoring instrument to perform high-frequency measurements. S4: Multi-source data calculation and deviation analysis: The platform automatically calls the centerline deviation calculation module of the corresponding alignment to calculate the current deviation; calculates the cumulative value and rate of change of vertical and horizontal displacement of each monitoring point of the steel support; calculates the deflection value of the guide beam and deducts the elevation change correction value caused by the climbing or descending of the beam. S5: Dual-control collaborative early warning and information feedback: When the cumulative value or rate of change of any monitoring item reaches the preset early warning threshold, the platform automatically pushes early warning information through various means and can retrieve real-time video images of the early warning point for verification.

2. The monitoring method according to claim 1, characterized in that, S1 specifically includes the following steps: S11. Control point layout: At least 3 control point prisms shall be laid out on the stable concrete structure outside the construction influence zone. The control points shall be arranged in an equilateral or isosceles triangle, and all monitoring points shall be located within the coverage area of ​​the triangle formed by the control points, following the principle of controlling the short side with the long side and the best frontal angle; if there is no existing stable structure, a 40cm×40cm×1.2m concrete pier shall be poured as the foundation for the control points. S12. Equipment Setup and Baseline Acquisition: The AI ​​video monitoring measuring instrument will be set up in a safe location with a line of sight to all control points and monitoring areas, and fixed with a 2m aluminum alloy tower. If the ground is not hardened, a 1m×1m×50cm concrete foundation with a depth of 30cm will be poured. The coordinates of all control points in the design coordinate system provided by the construction unit will be obtained. S13. Coordinate system transformation and accuracy verification: Control the AI ​​video monitoring measuring instrument to measure the coordinates of each control point in sequence, use the Bursa seven-parameter transformation model or the least squares method to calculate and fix the transformation parameters between the equipment coordinate system and the design coordinate system; measure the coordinates of one independent check point, and if the three-dimensional coordinate deviation of the check point is ≤ ±2mm, the coordinate system is deemed to be aligned; otherwise, the control points are re-laid out and the above steps are repeated. S14. Reference calibration mechanism setting: The system automatically sets all control point coordinates to be checked every 30 minutes. If the control point deviation exceeds the preset threshold, the system automatically recalculates the conversion parameters and calibrates the reference of all monitoring points to eliminate reference drift caused by factors such as temperature deformation and equipment micro-motion.

3. The monitoring method according to claim 1, characterized in that, S2 specifically includes the following steps: S21. Layout of steel support deformation monitoring points: On the front of the columns and key cross braces of the steel support to be monitored, attach 5cm×5cm high reflective patches; if night construction is required, replace them with welded and fixed right-angle prisms; each steel support shall have no less than 2 monitoring points. S22. Layout of centerline deviation monitoring points under different working conditions: For the straight-line jacking case: At least one checkerboard target is set up at the front, middle and rear of the beam in the jacking direction, for a total of ≥3 non-collinear targets, for global inverse algorithm calculation; at the same time, one checkerboard target is set up near the centerline of the front and rear cross sections of the beam for direct deviation method calculation and result cross-checking. Curved jacking case: Set up a checkerboard target at each of the four corner points of the jacking beam segment; receive the theoretical design coordinates corresponding to each jacking stroke of the four corner points of the beam segment exported by the construction unit based on the BIM model, and store them in the platform database; S23. Deflection monitoring point layout of guide beam: Weld a large deflection monitoring prism at the upper edge or centroid of the front end of the guide beam; no less than 4 points shall be laid out on the straight section and no less than 2 points shall be laid out on the curved section. S24. Initial value acquisition and filing: Generate automated monitoring tasks through the platform and set the patrol route; control the AI ​​video monitoring measuring instrument to continuously and independently measure all monitoring points 3 times, take the average value of the stable values ​​as the initial three-dimensional coordinates of each monitoring point, store them in the platform and complete the filing.

4. The monitoring method according to claim 1, characterized in that, S3 specifically includes the following steps: S31. Adaptive monitoring frequency strategy: Steel support deformation and centerline deviation monitoring: frequency is 1 time / hour when the jacking is stationary; automatically increased to 1 time / 5 minutes during jacking operation; further increased to 1 time / 2 minutes when an early warning is issued; Guide beam deflection monitoring: The entire process is continuously acquired at an equivalent rate of 100 frames / second, and the deflection time history curve is output in real time; Panoramic Image Acquisition: A panoramic image of the monitoring area is automatically acquired and archived once a day at 12 noon for the purpose of visual traceability of the construction process; S32. Automatic measurement of routine monitoring points: For steel support deformation monitoring points and centerline deviation monitoring points, the AI ​​video monitoring measuring instrument strictly follows the preset cruise route, rotates to each point in sequence, and executes the automated process of AI target center recognition → laser aiming and distance measurement → three-dimensional coordinate calculation → data upload. S33. Intelligent Tracking of Moving Deflection Monitoring Points: For deflection monitoring points that move forward with the beam, the following autonomous tracking process is executed: First, measure the current coordinates of at least three non-collinear checkerboard targets on the same rigid body; Solve for the optimal rigid body transformation (i.e., rotation matrix R and translation vector T) from the initial point set to the current point set. Rigid body transformation is applied to the initial coordinates of the deflection monitoring target to obtain the basic estimated coordinates; combined with the current jacking speed and monitoring cycle, the target movement compensation is calculated to further narrow the estimated spatial range. The bridge deflection visual monitoring instrument is controlled within a narrowed range. Using the YOLO target detection algorithm combined with sub-pixel corner point extraction technology, it accurately locates the center of the deflection monitoring prism and completes high-frequency measurement.

5. The monitoring method according to claim 1, characterized in that, S4 specifically includes the following steps: S41. Steel support deformation calculation: Calculate the difference between the current coordinates and the initial coordinates of each monitoring point, and obtain the cumulative value of vertical displacement and horizontal displacement and the rate of change per unit time. S42. Calculation of axis deviation under different working conditions: Linear jacking operation: Simultaneously call the dual-mode solution module to output the calculation results of the global inverse algorithm and the direct deviation method respectively; when the deviation between the two results is ≤2mm, the maximum value is taken as the final centerline deviation; when the deviation is >2mm, the monitoring frequency is automatically encrypted and an abnormal prompt is triggered. Curved jacking condition: The corner point comparison module is called to compare the real-time measured coordinates of the four corner points with the theoretical design coordinates corresponding to the current jacking stroke point by point, and calculate the northward deviation ΔN and eastward deviation ΔE of each corner point; by synthesizing the deviation vectors of the four corner points, the overall posture and centerline deviation of the curved beam segment are comprehensively evaluated. S43. Guide Beam Deflection Calculation and Correction: Calculates the difference between the current vertical coordinates and the initial vertical coordinates of the deflection monitoring target; automatically deducts elevation changes caused by beam climbing or descending; correction methods can be switched. Average slope correction method: Calculate the average slope of elevation change based on the difference in design elevation before and after the beam is launched and the total launching distance, and obtain the correction value by combining it with the current launching distance; Step-by-step theoretical correction method: Import the theoretical elevation data of the beam body provided by the construction unit for each step of the push-up stroke, and directly use this data for correction; the final guide beam deflection calculation formula is: guide beam deflection = current vertical coordinate - initial vertical coordinate - beam body elevation correction value.

6. The monitoring method according to claim 1, characterized in that, S5 specifically includes the following steps: S51. Warning Threshold Setting: The platform uniformly manages the control thresholds of all monitoring items. The yellow warning value is 70% of the control value, and the red warning value is 100% of the control value. Among them, the vertical and horizontal displacement of the steel support adopts dual control of cumulative value + 2mm / h change rate, while the centerline deviation and guide beam deflection adopt single control of cumulative value. S52. Tiered Early Warning and Response: Yellow Alert: When any monitoring item reaches the yellow alert threshold, the platform sends an SMS reminder to the project manager, and the interface highlights the alert point, value, and trend; it automatically increases the monitoring frequency of the corresponding monitoring item and strengthens remote video surveillance; Red Alert: When any monitoring item reaches the red alert threshold, the on-site audible and visual alarm will be triggered immediately; at the same time, alarm information and real-time image snapshots of the warning point will be pushed to all relevant parties such as the project manager, supervision unit, and construction unit; the platform will automatically generate a written early warning report containing monitoring data curves and deformation analysis within 2 hours. S53. Information Feedback and Data Management: The platform displays data curves, deformation cloud maps, and equipment operating status for each monitoring item in real time; it supports AI comparison between the original image and the current image of the warning point to intuitively display the deformation situation; it automatically generates daily, weekly, and periodic monitoring reports, and all monitoring data and reports are permanently archived and can be exported and viewed at any time.

7. The monitoring system used in the multi-parameter monitoring method for bridge jacking construction based on vision and laser fusion as described in any one of claims 1-6, characterized in that, include: Front-end intelligent sensing unit: includes at least one set of AI video monitoring and measuring instrument and one set of bridge deflection visual monitoring instrument; the AI ​​video monitoring and measuring instrument integrates a high-precision gimbal, laser ranging module and AI video recognition module, used to automatically search, aim and measure the three-dimensional coordinates of preset cooperative targets; the bridge deflection visual monitoring instrument integrates an industrial-grade fixed-focus camera and a high-frequency image acquisition module, used to perform sub-pixel level displacement measurement of cooperative targets at the front end of the guide beam; Multi-type cooperative target units include: control point large prisms set on stable benchmark points, reflective patches or small prisms set on steel support deformation monitoring points, checkerboard targets set on beams, and deflection monitoring large prisms set on the front end of guide beams. Automated monitoring, control and calculation platform: The platform receives measurement data from the front-end intelligent sensing unit and has built-in modules for automatic coordinate system alignment and periodic verification, dual-mode calculation of straight jacking centerline deviation, corner point comparison of curved jacking centerline deviation, deflection monitoring target position estimation and AI repositioning based on rigid body kinematics, guide beam deflection elevation correction, adaptive monitoring frequency adjustment, and multi-parameter collaborative early warning based on dual control of cumulative value and rate of change. The dual-mode solution module for the deviation of the centerline of the linear jacking push includes: Global inverse algorithm calculation unit: Based on the real-time measured coordinates of at least 3 non-collinear checkerboard targets on the beam, solve for the optimal rotation and translation transformation matrix of the beam relative to the initial state at the current moment; apply the transformation matrix to 2 pre-set feature points on the central axis of the beam to obtain their current actual coordinates; calculate the vertical distance from the two feature points to the theoretical central axis, and take the maximum value as the current deviation of the central axis of the beam; Direct deviation method calculation unit: At least one checkerboard target is set up on each side of different cross sections of the beam along the jacking direction, and the vertical distance from each target to the theoretical centerline is calculated. The distance calculated in real time is compared with the theoretical distance measured at the initial installation, and the change is the centerline deviation at the corresponding cross section. The dual-mode solution module outputs the calculation results of the two methods and performs cross-checking. When the deviation between the calculation results of the two methods exceeds 2mm, the system automatically encrypts the monitoring frequency and triggers an abnormality prompt. The curve jacking centerline deviation corner point comparison module addresses the issue of the high complexity of directly calculating the vertical distance when the centerline of the curve jacking is an arc. It adopts a discrete point pose comparison method: it receives the stroke-by-stroke theoretical design coordinates of the four corner points of the jacked beam segment exported by the construction unit based on the BIM model, and automatically updates the theoretical coordinates after each jacking stroke is completed; it monitors the actual spatial coordinates of the checkerboard targets placed at the four corner points of the beam segment in real time, and calculates the northward deviation ΔN and eastward deviation ΔE of each corner point; and it comprehensively evaluates the overall pose and centerline deviation status of the curve beam segment by synthesizing the deviation vectors of the four corner points. The deflection monitoring target position estimation and AI relocalization module based on rigid body kinematics performs the following steps: During the initialization phase, the initial coordinates of the large prism P0 for deflection monitoring at the front end of the guide beam, as well as the initial coordinates of at least three non-collinear checkerboard targets (Q1, Q2, Q3...) located on the same beam or guide beam, are measured and recorded. Periodic monitoring phase: First, measure the updated coordinates (Q1', Q2', Q3'...) of the above chessboard target at the current moment; Rigid body transformation solution: Solve for the optimal rigid body transformation from the initial point set {Q} to the current point set {Q'} to obtain the rotation matrix R and the translation vector T; Dual position estimation: Apply rigid body transformation to the initial coordinate P0 of the deflection monitoring target to obtain the basic estimated coordinate P_pre=R・P0+T; Combine the current pushing speed and monitoring cycle to calculate the compensation amount generated by the target movement, and further narrow the estimated spatial coordinate range. AI-powered intelligent targeting: The bridge deflection visual monitoring instrument is controlled to accurately identify and lock the center of the deflection monitoring prism within the aforementioned narrowed range using the YOLO target detection algorithm combined with sub-pixel corner extraction technology, thus completing high-precision coordinate measurement.

8. The monitoring system according to claim 7, characterized in that, The AI ​​video surveillance measuring instrument has the following characteristics: The high-precision gimbal has a horizontal motion range of 0° to 360° and a vertical motion range of -90° to +90°. The camera has an image resolution of 1920×1080px and features optical zoom of more than 32x. Automatic monitoring accuracy: ±0.5mm (measuring range ≤50m), ±1mm (measuring range 50m~100m); Supports remote visual deployment of monitoring points, automatic cruise measurement, automatic target tracking and aiming; automatically checks the coordinates of the control point prism every 30 minutes and calibrates the measurement benchmark. The bridge deflection visual monitoring instrument has the following characteristics: It uses an industrial-grade fixed-focus lens with an image resolution of 4096×2460px; Measurement accuracy: 0.1mm / 10m, 0.2mm / 20m, 0.3mm / 30m; The physical frame rate is 25 frames per second, and displacement monitoring is achieved at an equivalent of 100 frames per second through frame interpolation technology. It supports simultaneous real-time monitoring of up to 7 monitoring points.

9. The monitoring system according to claim 7, characterized in that, The checkerboard target is a 240mm×240mm square black and white checkerboard, with prisms integrated on its back or side; the target is mounted on the beam via a special support and universal joint, allowing for adjustable spatial angle.

Images